History of the Stanford-Binet Intelligence Scales: Content and Psychometrics

Kirk A. Becker

The 2003 publication of the Stanford-Binet IntelligenceScales, Fifth Edition represents the latest in a series ofinnovations in the assessment of intelligence and abilities.Using qualitative and quantitative methods, this bulletinexamines the similarities and differences between the differenteditions of the Stanford-Binet published over the past century.It discusses the development and integration of age-scale andpoint-scale formats for subtests, the theoretical structure of thetest (single versus hierarchical, use of Nonverbal and Verbaldomains, and links to the Cattell-Horn-Carroll theory), andchanges in item content related to this theoretical structure.The new edition provides greater differentiation in themeasurement of abilities, the precursors of which werenevertheless present in earlier editions.

Stanford-Binet Intelligence Scales, Fifth EditionAssessment Service Bulletin Number 1

Copyright 2003 by The Riverside Publishing Company. All rights reserved. Permission is hereby granted tophotocopy the pages of this booklet. These copies may not be sold and further distribution is expressly prohibited. Except as authorized above, prior written permission must be obtained from the Riverside PublishingCompany to reproduce or transmit this work or portions thereof, in any other form or by any other electronic ormechanical means, including any information storage or retrieval system, unless such copying is expressly permitted by federal copyright law. Address inquiries to Contracts and Permissions Department, The RiversidePublishing Company, 425 Spring Lake Drive, Itasca, IL 60143-2079.

Printed in the United States of America.

Reference Citation

To cite this document, use:

Becker, K. A. (2003). History of the Stanford-Binet intelligence scales: Content and psychometrics.(Stanford-Binet Intelligence Scales, Fifth Edition Assessment Service Bulletin No. 1). Itasca, IL: Riverside Publishing.

For technical information, please call 1.800.323.9540, visit our website at www.stanford-binet.com, or e-mail usat rpcwebmaster@hmco.com

1 2 3 4 5 6 XBS 06 05 04 03

1History of the Stanford-Binet Intelligence Scales:Content and Psychometrics

Overview

The study of the history of major psychological assessment instruments is hardlyan active field. Most professionals view assessment instruments simply aspractical tools of little interest beyond their immediate applied use. Nevertheless,when authors begin a revision of such an instrument, particularly when theinstrument has already undergone a number of revisions, they are wise toexamine the history of that assessment to provide continuity of measurement, toimprove on the features, and to overcome limitations of earlier versions. Inaddition, understanding the history of a test may help the clinician compare thescores on the newest edition of the test to the scores on an earlier edition of thetest with which they are familiar.

The history of assessment instruments may also be of interest to historians ofeducation, psychology, and science. Such interest would probably focus on anoriginal approach to assessment and the subsequent development of thatapproach. Truly original assessment tools are rare. Such instruments, and theirattendant theories, may have a sufficiently large impact on the field as toessentially spur a revolution in theory and application, along the lines of Kuhns(1970) discussion of paradigm shifts in the history of science. A new assessmentparadigm would begin with revolutionary fervor but would cool over time intonormal science. The history of intelligence measurement, and particularlyBinets major contribution, almost certainly qualifies as an original approach toassessment. As with other scientific revolutions described by Kuhn, Binets initialcontribution led to a period of revolutionary science followed by a drawn-outperiod of normal science, during which the original brilliant insight was improvedgradually over time with steady and practical enhancements. Kuhn noted thatalthough revolutionary science tends to grab the attention and the spotlight,normal science is what tends to be responsible for real progress, in both basicscience and its applications. One can probably say the same for the history ofintelligence tests and intelligence testing.

This bulletin provides a history of the Stanford-Binet, beginning with itsprecursors in the work of Alfred Binet and Theodore Simon at the turn of the20th century. It focuses on its five American editions, from the first version, whichwas published in 1916, to the most recent version, the Stanford-Binet IntelligenceScales, Fifth Edition (SB5) (Roid, 2003a), published in 2003. Although the purposeof this bulletin is to describe the similarities and differences across the fiveAmerican editions, readers with a broader interest in the history of assessmentinstruments may also find the discussion of interest, although it is not theprimary goal of this bulletin to provide the full history of intelligence testing interms of a Kuhnian scientific revolution.

2History of the Stanford-Binet

At a conference in Rome in April 1905, Dr. Henri Beaunis read a paper preparedby Alfred Binet and Theodore Simon that announced the development of anobjective measure capable of diagnosing different degrees of mental retardation(Wolf, 1973). This announcement was followed 2 months later by the publicationof the Binet-Simon Intelligence Test in LAne Psychologique (Binet & Simon,1905). The original form of this test was expanded and revised, leading to newversions in 1908 and 1911. The new forms were the result of extensive researchand testing involving normal as well as mentally retarded examinees.

In 1916, Lewis Terman authored The Measurement of Intelligence: AnExplanation of and a Complete Guide for the Use of the Stanford Revision andExtension of the Binet-Simon Intelligence Scale (Terman, 1916). This manualpresented translations and adaptations of the French items, as well as new itemsthat Terman had developed and tested between 1904 and 1915. Although therewere other translations of the Binet-Simon available around this time (Binet &Simon, 1916; Kuhlmann, 1912; Melville, 1917; Herring, 1922), Termans normativestudies and his methodical approach are credited with the success of theStanford-Binet (Minton, 1988).

Over the two decades following the initial publication of the Stanford-Binet,Terman continued his research and development of the test. Working with MaudMerrill, first his student and later a fellow professor and research collaborator atStanford University, Terman created two parallel forms of the Stanford-Binet.These forms used many of the items from the original Stanford revision and addeda substantial number of new items. With regard to this revision, Terman andMerrill wrote that they had provided two scales instead of one, have extendedthem so as to afford a more adequate sampling of abilities at the upper and lowerlevels, have defined still more meticulously the procedures for administration andscoring, and have based the standardization upon larger and more representativepopulations (Terman & Merrill, 1937, p. ix). These parallel forms were publishedas Form L (for Lewis) and Form M (for Maud) of the Stanford-Binet.

In the 1950s, Merrill took the lead in revising the Stanford-Binet, selecting thebest items from Forms L and M to include in a new version of the test. The twoforms from 1937 were combined to create the Form L-M. This form was publishedin 1960 (Terman & Merrill, 1960) and was later renormed in 1973 (Terman &Merrill, 1973). This form added alternate items at all levels, but otherwise, theformat remained similar to the 1937 forms.

The Stanford-Binet Intelligence Scale: Fourth Edition (Thorndike, Hagen, &Sattler, 1986) moved from the age-scale format introduced by Binet to a point-scaleformat (the section on Test Structure that follows provides details on thecharacteristics of age and point scales). Many of the items and item-types from theprior editions were included in the Fourth Edition, and extended scales were createdusing the same types of items and activities. In the Absurdities test, for example,four classic items were used in addition to 28 new items. Also, several completelynew subtests, such as Matrices and Equation Building, were created. Besides thenew and expanded tests, the Fourth Edition provided several factors (VerbalReasoning, Abstract/Visual Reasoning, Quantitative Reasoning, and Short-TermMemory) in addition to IQ. Although the prior versions had items that related to

3these four areas (McNemar, 1942; Sattler, 1965), the published test had never offered scores for these factors. The Fourth Edition also formalized the practice ofmulti-stage testing, in which performance on the Vocabulary scale determines thestarting point for subsequent tests. While some examiners used the vocabulary testfor routing on earlier editions of the test, this was not official practice.

In 2003, the Fifth Edition (Roid, 2003a) was published. This edition attempts tocarry on the tradition of the prior editions while taking advantage of currentresearch in measurement and cognitive abilities. Like the Fourth Edition, theSB5 includes multiple factors. These factors are modified from those on theFourth Edition, but represent abilities assessed by all former versions of the test.The use of routing subtests continues, with a nonverbal routing test added tocomplement vocabulary. The Fifth Edition reintroduces the age-scale format forthe body of the test, presenting a variety of items at each level of the test. Theage-scale is intended to provide a variety of content to keep examinees involvedin the testing experience and to allow for the introduction of developmentallydistinct items across levels.

Test Structure

The Stanford-Binet is one of the first examples of an adaptive test (Reckase,1989). Examiners use the information they have about an examinee to determinewhere to begin testing and administer only those items that are appropriate forthat examinee. This format reduces the time required to obtain reliableinformation from a test and decreases the frustration examinees experience whenpresented with items that are too hard or too easy. The use of multiple possiblestarting points, along with basal and ceiling rules, limits the time required toadminister the test and maximizes the information obtained from each item.

One element of test structure that appears throughout the history of theStanford-Binet is that of point scales and age scales. A point scale is the currentlywidespread arrangement of tests into subtests, with all items of a given typeadministered together. Age scales, long a part of the Stanford-Binet format, maynot be familiar to the current generation of examiners. Initially, this format wasused to provide a direct translation of the childs performance to mental age.Psychometric as well as developmental information was used to place items onthe test. Examinees experienced a variety of items that changed bothquantitatively and qualitatively. Definitions, for example, went from concretewords to abstract words to the comparison of abstract words. The argument hasoften been made that this format is more engaging and provides a richeropportunity for the examiner to observe the examinees performance. Althoughthis perspective guided the development of the Fifth Edition, it is not without itsdetractors. With the publication of the 1916 edition, Robert Yerkes began a seriesof debates with Lewis Terman on the appropriateness of the age scale (Yerkes,1917). While Termans methods prevailed at the time, the structure of manycurrent tests (e.g., Wechsler, 1991) shows the popularity of the point-scale subtest.

Throughout most of its history, the Stanford-Binet maintained a hybrid structure,combining point-scale and age-scale formats. Terman presented two parallelvocabulary scales in 1916, and every version of the Stanford-Binet except Form Mhas included a vocabulary scale. In 1986, the Fourth Edition provided a standard

method for using Vocabulary as a routing test to determine where to begin testing.Although this was the first formal mention of routing in a Stanford-Binet manual,using vocabulary for this purpose had been the unofficial practice of manyexaminers for decades. The Fifth Edition includes a nonverbal routing test inaddition to Vocabulary, and it uses performance on these subtests to route to theNonverbal and Verbal age scales. Table 1 provides a summary of the structure andfeatures of the different editions of the Stanford-Binet.

Table 1Test Structure of the Stanford-Binet: 1916 to 2003Edition Structure Abilities Measured1916 Parallel vocabulary tests General intelligence

Single age scale1937 Form L vocabulary test General intelligence

Parallel age scales1960/1973 Vocabulary test General intelligence

Single age scale1986 Vocabulary routing test General intelligence

Subtest point scales Verbal Reasoning Abstract/Visual Reasoning Quantitative Reasoning Short-Term Memory

2003 Hybrid structure General intelligence Verbal routing test Knowledge Nonverbal routing test Fluid Reasoning Verbal and nonverbal age scales Quantitative Reasoning

Visual-Spatial Processing Working Memory Nonverbal IQ Verbal IQ

Verbal and Nonverbal Content

The verbal content of the Stanford-Binet has been of concern since the firstpublication in 1916, in which verbal items predominated. Subsequent revisionshave attempted to better balance the verbal and nonverbal items, and McNemarreports that attempts were made to create a parallel nonverbal form in 1937(McNemar, 1942). During the 1986 revision, the authors tried to create anonverbal routing test using Matrices; however, sufficient low-end items were notcreated. While several nonverbal forms were recommended for different editions ofthe Stanford-Binet (e.g., McNemar, 1942; Delaney & Hopkins, 1987; Glaub &Kamphaus, 1991), the test never had a published nonverbal form prior to 2003.The Fifth Edition provides an equal balance of verbal and nonverbal contentwithin each factor. Norms are provided for Verbal IQ (VIQ) and Nonverbal IQ(NVIQ) scores as well as for Full Scale IQ (FSIQ), and the correlations betweenthese scores range from .94 to .97 across the age range of the test. Tables 2 and 3present the nonverbal content of all editions of the Stanford-Binet. Table 4compares the VIQ and NVIQ of the Fifth Edition with FSIQ.

4

5Table 2Nonverbal Items on the Stanford-Binet: 1916 to 1973Level 1916 Form L Form M Form L-M2-0 n/a 2 2 32-6 n/a 1 2 13-0 0 5 3 63-6 n/a 2 5 54-0 3 2 2 14-6 n/a 2 3 25 3 4 4 66 0 5 2 27 2 2 1 28 1 0 0 09 1 2 1 110 2 1 0 011 n/a 1 1 112 1 0 2 213 n/a 3 1 314 0 1 1 0AA 0 0 1 1SAI 1 0 0 0SAII n/a 0 0 0SAIII n/a 1 0 0Total NV items 14 34 31 36Total items 90 129 129 142Total percent 16% 26% 24% 25%

Note. The 1916 edition did not have all of the levels that later editions had, so the levels not applicable to thatedition are marked n/a.Abbreviations: AA = average adult, SAI = superior adult I, SAII = superior adult II, SAIII = superior adult III, NV = nonverbal

Brief/Abbreviated Forms

Testing time and examinee fatigue have always been important issues forpsychological assessment practitioners. Recognizing that testing time might needto be reduced, Lewis Terman included instructions for a minimal level of leniencein establishing basal and ceiling performance (Terman, 1916). An examinee, hereasoned, could miss one item at an age level and still have a basal. Later, in1937, instructions were included with the Stanford-Binet for the administration

Table 3Nonverbal Content of the Fourth and Fifth Editions of the Stanford-BinetSB IV SB5Bead Memory Nonverbal Fluid Reasoning (Object Series/Matrices)Pattern Analysis Nonverbal Knowledge (Procedural Knowledge, Picture Absurdities)Absurdities Nonverbal Quantitative Reasoning (Quantitative Reasoning)Copying Nonverbal Visual-Spatial Processing (Form Board, Form Patterns)Memory for Objects Nonverbal Working Memory (Delayed Response, Block Span)MatricesPaper Folding and Cutting

6Table 4Correlations of Verbal IQ (VIQ) and Nonverbal IQ (NVIQ) With FullScale IQ (FSIQ)

2-0 to 5-11 6-0 to 10-11 11-0 to 16-11 17-0 to 50-11 51 to 85+ 2 to 85+VIQ/FSIQ .96 .96 .96 .97 .97 .96NVIQ/FSIQ .94 .95 .96 .97 .97 .96

Note. From Roid, 2003a.

of an abbreviated test battery (Terman & Merrill, 1937). With little loss of reliability, designated items could be omitted from the test to conserve time.Remaining items administered in this fashion were given more weight in thecalculation of IQ. This feature was retained in the 1960/1973 edition where, forinstance, an examiner could omit Memory for Sentences and Copying a BeadChain from Memory at age 13. The Fourth Edition also allowed for anabbreviated administration through its flexible selection of subtests. Differentcombinations of subtests were recommended for different purposes, including afour-test screening battery and a six-test screening battery. While these screenersdo not allow for a good measure of an individuals pattern of abilities, theyprovide a quick measure of IQ (Thorndike, Hagen, & Sattler, 1986). A similarmethod is used in the Fifth Edition, with Vocabulary and Object-Series/Matricesproviding the Abbreviated Battery IQ (ABIQ). The ABIQ, like the FSIQ, is equallyweighted in terms of verbal and nonverbal content. Table 5 summarizes thedifferent methods for abbreviated test administration over time.

Table 5Summary of Stanford-Binet Abbreviated FormsEdition Abbreviated Form1916 Leniency in basal/ceiling rules to save time1937 Selected items omitted1960/1973 Selected items omitted1986 4- and 6-subtest abbreviated batteries recommended2003 2-subtest Abbreviated Battery IQ available

Reviews of the Stanford-Binet

Along with the extensive research literature on the Stanford-Binet, reviews of thetest have been available since before the first Mental Measurement Yearbook waspublished (Buros, 1938; Pratt, 1917). The content of these reviews has beendistilled where possible in Table 6 to offer a summary of the advantages andpossible limitations of the different versions of the test. The 1916 edition of theStanford-Binet provided continuity with the Binet-Simon test, while at the sametime expanding the range of items and providing a large research sample. Thetest offered only limited nonverbal content, provided only sparse instructions onscoring some items, and was not an adequate measure of adult intelligence. The1937 editions again extended the range of items, provided a set of toys andobjects to engage young children, and added more nonverbal content,predominantly at the lower end of the scale. This edition also improved thepsychometric characteristics of the test by introducing a parallel form and more

7Table 6Features and Possible Limitations of the Stanford-Binet Over TimeYear Advantages Limitations1916 Contains alternate items at most age levels Inadequately measures adult mental capacity

Shares items to maintain continuity with earlier versions Has inadequate scoring and administrative procedures Emphasizes abstraction and novel problem solving at some points Extends range of items relative to Binet-Simon Measures only single factor (g) Based on extensive research literature Has nonuniform IQ standard deviation Extensive standardization performed Has single test form

Is verbally loaded1937 Contains alternate items at most levels Some items have ambiguous scoring rules

Shares items to maintain continuity with earlier versions Form M lacks vocabulary Extends range of items Has longer administration time than 1916 version Based on extensive research literature Measures only single factor (g) Contains more performance tests at earlier age levels Has nonuniform IQ standard deviation Contains more representative norms IQs not comparable across ages Includes parallel form Sample had higher SES and higher percentage of urban Uses toys to make test more engaging for young children children than general population Verbal items allow subjects to display fluency, imagination, Has unequal coverage of different abilities at

unusual or advanced concepts, and complex linguistic usage different levels Is verbally loaded

1960/1973 Administers several varied tests to each examinee to keep Has inadequate ceiling for adolescents and highly children interested gifted examinees

Retains best items from Forms L and M Measures only single factor (g) Has better layout than previous versions Separates scoring standards from items Manual presents clear scoring rules Is verbally loaded Contains alternate items at each age level Shares items to maintain continuity with earlier versions Eliminates items that are no longer appropriate Based on extensive research literature Presents stimulus material in spiral-bound book Has uniform IQ standard deviation Uses toys to make test more engaging for young children

1986 Contains both a general composite score and several Less gamelike than earlier versions; yields less factor scores information from styles and strategies due to decreased

Shares items to maintain continuity with earlier versions examiner/examinee interaction Easel format with directions, scoring criteria, and stimuli Contains no toys

makes administration easier Norming sample overrepresents managerial/professional Emphasizes abstraction and novel problem solving; and college-educated adults and their children

emphasizes verbal reasoning less compared with prior versions Has possible lack of comparability in the content of area Technical Manual reports extensive validity studies scores at different ages due to variability of subtests Has flexible administration procedures used in their computation Contains higher ceilings for advanced adolescents than Has psychometric rather than developmental emphasis

Form L-M Has standard deviation of 16 rather than 15 for Number of basic concepts in preschool level tests compares composite scores; M = 50, SD = 8 for subtests

favorably with other tests for that age range Contains subjectivity (examiner preference) when Contains understandable age-level instructions for young determining subtests used to compute composite score

children Unable to diagnose mild retardation before age 4 and Uses adaptive testing (routing) to economize on moderate retardation before age 5

administration time and reduce examinee frustration Uses explicit theoretical framework as guide for item

development and alignment of subtests within modeled hierarchy Has wider age range than prior versions (2-0 through 23) Creatively extends many classic item types

Table 6 (Continued)Features and Possible Limitations of the Stanford-Binet Over TimeYear Advantages2003 More gamelike than earlier versions with colorful artwork, toys, and manipulatives

Matches norms to 2000 U.S. Census Contains nonverbal as well as verbal routing test Contains both a general composite score and several factor scores Shares items to maintain continuity with earlier versions Covers age range of 2-0 through 85+ Change-sensitive scores allow for evaluation of extreme performance Has easel format with directions, scoring criteria, and stimuli, for easy administration Has equal balance of verbal and nonverbal content in all factors Contains Nonverbal IQ Has standard deviation of 15 for composite scores, allowing easy comparison with other tests; M = 10, SD = 3 for subtests Uses adaptive testing (routing) to economize on administration time and reduce examinee frustration Uses explicit theoretical framework as guide for item development and alignment of subtests within modeled hierarchy Extends low-end items, allowing earlier identification of individuals with delays or cognitive difficulties Extends high-end items to measure gifted adolescents and adults

Note. Because the 2003 measure was just published, a discussion of possible limitations does not yet exist in the literature. (See discussion in text.)

8

representative norms. While better than the 1916 edition, Forms L and M werestill criticized for the quality of the scoring rules, the paucity of nonverbal contentat the upper levels of the test, and the nonuniform standard deviation of IQ thatled to different interpretations of IQ at different ages. This last point wascorrected with the publication of the Form L-M, which provided tables to correctfor this. Not only were the best items from the parallel 1937 forms used to createForm L-M, but also the ambiguities in scoring were cleared up, and the stimulusmaterials were presented in a much more convenient bound format. Somereviewers suggested that the format could be further improved by providing thescoring standards alongside the items in a single book. The test was alsocriticized for remaining heavily weighted with verbal materials, having what wasperceived as an inadequate ceiling for adolescents and highly gifted examinees,and continuing to provide only a single measure of general intelligence.

The Fourth Edition attempted to address many concerns that had been raisedwith prior versions of the test, while maintaining the same types of tasks anditems. In particular, this version of the test offered several factor scores based onan explicit theoretical framework. This test introduced the easel format to theStanford-Binet, providing both administration and scoring information as well asstimulus material in one place. The test featured higher ceilings for adolescentsand over five times as many nonverbal items as the previous edition. Althoughthe test provided for flexible administration, such a degree of flexibility can alsolead to unneeded complexity. The test introduced creative extensions of classicitems, but it lacked many of the toys and other interesting stimuli from theearlier editions. The continued use of a standard deviation of 16 was alsocriticized, as was the sample weighting done to approximate the characteristics ofthe general population.

The Fifth Edition is too new at this point to provide a list of possiblelimitations collected from the literature. Attempts were made to address thelimitations of prior versions, while maintaining the advantages. Artwork and

Table 8SB IV and SB5 Factor Correlations

Factor Correlations (Corrected for Restriction or Expansion of Variance) Factor Correlations (Uncorrected)

SB IV Scores SB IV ScoresSB5 Scores VR A/VR QR STM VR A/VR QR STM M SDKN .73 .57 108.64 9.85FR .54 .48 106.96 12.93VS .69 .64 107.90 13.20QR .77 .73 107.37 13.34WM .64 .55 104.33 11.76M 110.71 111.96 112.11 106.66 110.71 111.96 112.11 106.66SD 12.71 14.24 14.90 13.27 12.71 14.24 14.90 13.27

Note. N = 104. SB IV scores include Verbal Reasoning (VR), Abstract/Visual Reasoning (A/VR), Quantitative Reasoning(QR), and Short-Term Memory (STM). SB5 scores include Knowledge (KN), Fluid Reasoning (FR), Visual-Spatial Processing (VS), Quantitative Reasoning (QR), and Working Memory (WM).

manipulatives have been improved, and toys and gamelike materials wereincluded. The SB5 does not allow for as many administrative options as theFourth Edition; this makes for a more straightforward testing session. The FifthEdition covers the widest age range of any Stanford-Binet (2 through 85+ years)and addresses the criticism about verbal content, norms, and the standarddeviation. (The Fifth Edition uses a standard deviation of 15 for its IQ scales.)

Score/Test Comparability

With any newly revised test, questions arise about the relationship betweenscores and interpretations of the new version relative to the research and use of the prior editions. With the research tradition of the Stanford-Binet (as of 2003, over 2,200 articles), it is especially important to provide thesecomparisons. The comparability of new and old forms, as well as the validityevidence reported in the Technical Manual for the Fifth Edition (Roid, 2003b),provide the initial evidence for appropriate use of the test until independentresearch is published. Table 7 shows the correlations between general abilityon the last three editions of the Stanford-Binet, while Table 8 shows thecorrelations between factors on the Fourth and Fifth editions.

Table 7Correlations of FSIQ for the SB Form L-M, SB IV, and SB5

Test Correlations (Corrected for Restriction or Expansion of Variance) Test Correlations (Uncorrected)

SB5 SB IV SB5 SB IVSB IV 0.90 0.83Form L-M 0.85 0.80 0.80 0.81

Note. N = 104 for SB5 and SB IV, N = 80 for SB5 and SB L-M, N = 139 for SB IV and SB Form L-M

9

10

Test Comparability

In addition to the correlations between the SB5, the SB IV, and Form L-M,estimated equating tables were developed to show the relationship between scoreson these editions of the Stanford-Binet. The SB5 was administered incounterbalanced order to two samples; one sample was also administered the SB IV and the other Form L-M. Angoff s (1984) design II.A.1 was used to equatethe different forms of the test. Tables 9 and 10 show the results of these analyses,with the range of SB5 scores expected for given IQs from the SB IV and Form L-M.

Table 9Estimated Equating Table: Expected SB5 Full Scale IQ Ranges for SelectedSB IV Composite SAS Scores

SB IV SB5Composite SAS Score FSIQ

55 526070 677385 8285100 9699115 106113130 122128145 135143

Abbreviation: FSIQ = SB5 Full Scale IQ Score.

Table 10Estimated Equating Table: Expected SB5 Full Scale IQ Ranges for Selected SB Form L-M IQ Scores

SB Form L-M SB5IQ Scores FSIQ

55 637470 758285 8691100 96100115 105110130 114121145 122133

Shared Test Content

Comparisons between the exact content shared in the 1916 through 1973 versionsof the Stanford-Binet are somewhat difficult to quantify due to the occurrence ofitems at multiple levels (with different passing criteria). Moreover, many items onthe earlier forms of the test were grouped into testlets, and in some cases onlyparts of a testlet are shared across forms. To present a parsimonious descriptionof shared items over time, the percent of content shared with other forms ispresented for the SB IV and SB5, both of which present individual items rather

11

than groups of items. Interested readers are referred to the Form L-M manual fora complete content map of the 1937 and 1960/1973 forms of the test (Terman &Merrill, 1973). For both the Fourth and Fifth Editions, all test items were placedin a spreadsheet. Each item was checked against the content of all prior editions,and matching items were marked. Once all items were coded, the total number ofitems in each edition, as well as the total number of shared items overall, werecounted. Table 11 shows the results of this process.

Table 11Percent of SB IV and SB5 Items Appearing in Other Versions ofthe Stanford-Binet

SB5 SB IV L-M L M 1916 Binet Total

SB5Shared Items 42 27 19 20 11 2 69*Total Items 284 284 284 284 284 284 284Percent 14.8% 9.5% 6.7% 7.0% 3.9% 0.7% 24.3%

SB IVShared Items 42 32 28 14 14 0 75*Total Items 460 460 460 460 460 460 460Percent 9.1% 7.0% 6.1% 3.0% 3.0% 0.0% 16.3%

Note. *The total number of shared items is the number of unique items coming from prior editions. All itemsshared with the SB Form L-M, for example, would also have appeared on Form L and/or Form M.

Test Structure

While early versions of the Stanford-Binet yielded only a general intelligencescore, the items included a diverse mix of mental abilities. For example, items forage 4 of the 1916 edition contained visual-spatial (copying a square), quantitative(counting 4 pennies), knowledge (comprehension), and memory (repeating 4 digits)items. Although the types of items at any level of the 1916 to 1973 versions varied,almost all items on these forms relate to knowledge, fluid reasoning, visual-spatialprocessing, quantitative reasoning, or short-term/working memory. The FourthEdition separated the diverse types of items found on prior editions into subtestsand factors, grouping Visual-Spatial Processing and Fluid Reasoning together intoan Abstract/Visual Reasoning factor. The Fifth Edition also maintains these fivepredominant areas of mental ability with a factor structure that keeps FluidReasoning items separate from Visual-Spatial Processing items. The Fifth Editionalso places more emphasis on working memory compared with short-term memory.While working memory was present to some extent on earlier editions, it waslimited to Memory for Digits Reversed, with all other memory tasks involvingshort-term memory.

Table 12 presents a count of the types of items (using the Fifth Editioncategories) that appeared on prior versions of the Stanford-Binet. Theseclassifications are based on a content analysis of prior editions as well as onempirical studies (e.g., Woodcock, 1990). Of course, content analysis does notnecessarily ensure that an item will load on a particular factor in a factoranalysis, or that all practitioners will agree on the classifications given.

12

Additionally, a number of items and subtests fall across factors (for example,Knowledge and Visual-Spatial Processing contribute to Mutilated Pictures in theForm L-M). For the purpose of Table 12, items with multiple possibleclassifications are counted in both classes.

Table 12Item Content on the 1916 to 1973 Editions of the Stanford-Binet

SB5 SB IV Form L-M Form L Form M 1916Knowledge 26% 27% 40% 41% 38% 38%Fluid Reasoning 17% 14%* 20% 17% 21% 17%Visual-Spatial Processing 18% 17% 18% 16% 15% 11%Quantitative Reasoning 21% 17% 6% 6% 6% 8%Working Memory 12% 2% 4% 3% 4% 9%Short-Term Memory 6% 22% 9% 13% 11% 11%Other** 0% 0% 4% 3% 5% 6%

Note. *Verbal Relations is counted in Fluid Reasoning as well as Knowledge. Number Series is counted in FluidReasoning as well as Quantitative Reasoning.**These items cover such abilities as auditory processing, long-term retrieval, and judgements of weights.

Conclusions

The Stanford-Binet Intelligence Scales, Fifth Edition represents the latest in aseries of enhancements derived from the tradition of intelligence testingoriginated in 1905 by Alfred Binet and Theodore Simon. However, as Kuhn (1970)points out, the real progress in a science, and certainly the real progress in thefruits of a science, lies not in the revolutionary period but rather in the periods ofnormal science that follow it. According to Kuhn, the results gained in normalresearch are significant because they add to the scope and precision with whichthe paradigm can be applied (p. 36). The SB5 incorporates many insightsimplicitly designed into the early editions of the measure as implemented byBinet, Simon, Terman, and Merrill, but presents them in a way that provides vastpractical improvements in the areas of content coverage and psychometriccharacteristics. In this way, the revolutionary work of the earlier authors hasshaped the more recent enhancements and advancement of the test underThorndike, Hagen, and Sattler, and most recently, Roid.

13

References

Angoff, W. H. (1984). Scales, norms, and equivalent scores. Princeton, NJ: Educational TestingService.

Binet, A., & Simon, T. (1905). Mthodes nouvelles pour le diagnostic du niveau intellectual desanormaux. LAnne psychologique, 11, 191336.

Binet, A., & Simon, T. (1916). The development of intelligence in children (E. Kit, Trans.). Baltimore,MD: Williams & Wilkins.

Buros, O. K. (1938). The 1938 mental measurements yearbook. New Brunswick, NJ: RutgersUniversity Press.

Delaney, E. A., & Hopkins, T. F. (1987). Examiners handbook: An expanded guide for fourth editionusers. Itasca, IL: Riverside Publishing.

Glaub, V. E., & Kamphaus, R. W. (1991). Construction of a nonverbal adaptation of the Stanford-Binet Fourth Edition. Educational & Psychological Measurement, 51, 231241.

Herring, J. P. (1922). Herring revision of the Binet-Simon tests and verbal and abstract elements inintelligence examinations. New York: World Book Co.

Kuhlmann, F. (1912). A revision of the Binet-Simon system for measuring the intelligence ofchildren. Faribault, MN: Minnesota School for Feeble-Minded and Colony for Epileptics.

Kuhn, T. S. (1970). The structure of scientific revolutions (2nd ed.). Chicago: University of ChicagoPress.

McNemar, Q. (1942). The revision of the Stanford-Binet Scale: An analysis of the standardizationdata. New York: Houghton Mifflin Company.

Melville, N. J. (1917). Testing juvenile mentality. Philadelphia: J. B. Lippincott Company.

Minton, H. L. (1988). Lewis M. Terman: Pioneer in psychological testing. New York: New YorkUniversity Press.

Pratt, C. C. (1917). Book review: The measurement of intelligence. Journal of Applied Psychology, 1,191192.

Reckase, M. D. (1989). Adaptive testing: The evolution of a good idea. Educational Measurement:Issues & Practice, 8, 1115.

Roid, G. H. (2003a). Stanford-Binet Intelligence Scales, Fifth Edition. Itasca, IL: RiversidePublishing.

Roid, G. H. (2003b). Stanford-Binet Intelligence Scales, Fifth Edition: Technical Manual. Itasca, IL:Riverside Publishing.

Sattler, J. M. (1965). Analysis of functions of the 1960 Stanford-Binet Intelligence Scale, Form L-M.Journal of Clinical Psychology, 21, 173179.

Terman, L. M. (1916). The measurement of intelligence: An explanation of and a complete guide forthe use of the Stanford revision and extension of the Binet-Simon Intelligence Scale. Boston:Houghton Mifflin.

Terman, L. M., & Merrill, M. A. (1937). Measuring intelligence. Boston: Houghton Mifflin.

Terman, L. M., & Merrill, M. A. (1960). Stanford-Binet Intelligence Scale: Manual for the ThirdRevision Form L-M. Boston: Houghton Mifflin.

Terman, L. M., & Merrill, M. A. (1973). Stanford-Binet Intelligence Scale: Manual for the ThirdRevision Form L-M (1972 Norm Tables by R. L. Thorndike). Boston: Houghton Mifflin.

Thorndike, R. L., Hagen, E. P., & Sattler, J. M. (1986). Stanford-Binet Intelligence Scale: FourthEdition. Itasca, IL: Riverside Publishing.

Wechsler, D. (1991). The Wechsler Intelligence Scale for Children, Third Edition. San Antonio, TX:Psychological Corporation.

Wolf, T. H. (1973). Alfred Binet. Chicago: University of Chicago Press.

Woodcock, R. W. (1990). Theoretical foundations of the WJ-R measures of cognitive ability. Journalof Psychoeducational Assessment, 8, 231258.

Yerkes, R. M. (1917). The Binet versus the point scale method of measuring intelligence. Journal ofApplied Psychology, 1, 111122.

425 Spring Lake Drive

Itasca, IL 60143-2079

1.800.323.9540www.stanford-binet.com

9-96306